Augmenting Neuromuscular Disease Detection Using Optimally Parameterized Weighted Visibility Graph.

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Signal Processing

Background:

• Neuromuscular diseases like myopathy and amyotrophic lateral sclerosis (ALS) require accurate detection methods.
• Electromyography (EMG) signals are crucial for diagnosing neuromuscular disorders.
• Conventional weighted visibility graph (WVG) analysis of time series data, including EMG, faces challenges with noise and computational complexity.

Purpose of the Study:

• To propose a novel framework for neuromuscular disease detection using WVG analysis of EMG signals.
• To investigate the impact of WVG parameters (penetrable distance and scale factor) on signal analysis and computational efficiency.
• To enhance the discrimination capability between healthy individuals, myopathy patients, and ALS patients using optimized WVG parameters.

Main Methods:

• Electromyography (EMG) signals were analyzed using a weighted visibility graph (WVG) approach.
• The performance of WVG was evaluated by systematically varying the penetrable distance and scale factor parameters.
• Graph theory-based features were extracted, and classification was performed using optimal parameter values determined by ANOVA F-values.

Main Results:

• Optimizing WVG parameters significantly improved the discrimination capability between healthy, myopathy, and ALS EMG signals.
• High classification accuracies were achieved: 98.57% (healthy vs. myopathy), 98.09% (myopathy vs. ALS), 99.45% (healthy vs. ALS), and 99.05% (multi-class).
• Computational time was reduced by 96% with optimally selected WVG parameters compared to the traditional method.

Conclusions:

• The proposed WVG-aided framework with optimized parameters offers a highly accurate and computationally efficient method for neuromuscular disease detection from EMG signals.
• Adjusting penetrable distance and scale factor in WVG effectively mitigates noise sensitivity and reduces computational load.
• This approach demonstrates significant potential for clinical application in diagnosing myopathy and ALS.